Download Nature: Serine is a natural ligand and allosteric activator of pyruvate

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Signal transduction wikipedia , lookup

Polyclonal B cell response wikipedia , lookup

Cryobiology wikipedia , lookup

Catalytic triad wikipedia , lookup

Paracrine signalling wikipedia , lookup

Biochemical cascade wikipedia , lookup

Biosynthesis wikipedia , lookup

Specialized pro-resolving mediators wikipedia , lookup

Biochemistry wikipedia , lookup

Metabolism wikipedia , lookup

Amino acid synthesis wikipedia , lookup

Glycolysis wikipedia , lookup

Transcript
LETTER
doi:10.1038/nature11540
Serine is a natural ligand and allosteric activator of
pyruvate kinase M2
Barbara Chaneton1*, Petra Hillmann2*, Liang Zheng1, Agnès C. L. Martin2, Oliver D. K. Maddocks1,
Achuthanunni Chokkathukalam3, Joseph E. Coyle2, Andris Jankevics3,4, Finn P. Holding2, Karen H. Vousden1, Christian Frezza1{,
Marc O’Reilly2 & Eyal Gottlieb1
Cancer cells exhibit several unique metabolic phenotypes that are
critical for cell growth and proliferation1. Specifically, they overexpress the M2 isoform of the tightly regulated enzyme pyruvate
kinase (PKM2), which controls glycolytic flux, and are highly dependent on de novo biosynthesis of serine and glycine2. Here we describe
a new rheostat-like mechanistic relationship between PKM2 activity
and serine biosynthesis. We show that serine can bind to and activate
human PKM2, and that PKM2 activity in cells is reduced in response
to serine deprivation. This reduction in PKM2 activity shifts cells to a
fuel-efficient mode in which more pyruvate is diverted to the mitochondria and more glucose-derived carbon is channelled into serine
biosynthesis to support cell proliferation.
Metabolic fluxes in cancer cells are different from those in nontransformed cells1. In particular, a shift from oxidative phosphorylation
to aerobic glycolysis has been demonstrated, which is promoted by the
M2 isoform of pyruvate kinase3. PKM2 catalyses the final step of glycolysis, converting phosphoenolpyruvate (PEP) to pyruvate (Supplementary Fig. 1). Interestingly, PKM2, which is the predominant
isoform in cancer cells4,5, has low basal enzymatic activity compared
to the constitutively active splice variant PKM1 (ref. 6). Another metabolic pathway recently demonstrated to be crucial for cancer cell survival is the serine-biosynthesis pathway7–9. We investigated a potential
mechanistic link between the two pathways in cancer cells, whereby a
reduction in overall pyruvate kinase activity through the preferential
expression of PKM2 would cause the build-up of glycolytic intermediates for channelling into the serine-biosynthetic pathway. To test this
hypothesis, we used human colon carcinoma HCT116 cells, which
predominantly express the PKM2 isoform (Fig. 1a and Supplementary Fig. 2). Two discrete short hairpin RNA (shRNA) pools were used
to generate two independent HCT116-derived cell lines, shPKMa
and shPKMb, in which the expression of both the PKM1 and PKM2
isoforms was simultaneously and stably silenced (Fig. 1a and Supplementary Fig. 2c). Despite achieving a greater than 90% reduction
in PKM1 and PKM2 (hereafter termed PKM for simplicity) messenger
RNA and protein levels, compared to cells expressing non-targeting
shRNA (shCntrl), no compensatory transcriptional induction of the
PKL or PKR isoforms was observed in the shPKM cells (Supplementary Fig. 2). In line with this, liquid chromatography–mass spectrometry
(LC–MS) analysis of the steady-state levels of metabolites revealed a
100-fold increase in PEP concentration in shPKM cells accompanied
by a ,50% decrease in pyruvate levels, demonstrating a reduction in
intracellular pyruvate kinase activity (Fig. 1b). The stable silencing of
PKM in HCT116 cells did not alter cell proliferation rates or steady-state
levels of ATP (Fig. 1c, d). By contrast, the proliferation rates of HT29
and SW620 colon cancer cells were more sensitive to PKM silencing
(Supplementary Fig. 3a). Regardless of the effect on cellular proliferation
rates, PKM silencing universally increased the oxygen-consumption
rates by ,30%, with a corresponding decrease in extracellularacidification rates, indicators of increased oxidative phosphorylation
and decreased glycolysis, respectively (Supplementary Fig. 3b, c).
Because pyruvate kinase catalyses an important ATP-producing step
in glycolysis, the stability of intracellular ATP levels could be explained
by this compensatory increase in oxidative phosphorylation in response to PKM silencing. Thus, despite the predominant expression of
PKM2 in HCT116 cells, these cells still exhibit sufficient pyruvate
kinase activity to convert PEP to pyruvate and to facilitate aerobic
glycolysis.
Whereas PKM silencing caused a large increase in PEP concentration, pyruvate levels were decreased to a lesser extent (Fig. 1b). There
are several possible explanations for this. First, residual PKM could still
generate pyruvate, albeit at a lower rate. Second, pyruvate can be
synthesized from carbon sources other than glucose. Finally, pyruvate
can also be generated from PEP through a pyruvate-kinase-independent mechanism10, although this alternative pathway was not increased
in the knockdown cells (Supplementary Fig. 4).
In order to study the fate of glucose in PKM-inhibited cells, shCntrl
and shPKM cells were incubated in media containing uniformly 13Clabelled glucose (U-13C-glucose) and cells were extracted at different
time points. Several glucose-derived metabolites were tracked by
LC–MS (Fig. 2 and Supplementary Fig. 5), including pyruvate and
PEP. The ratio between these two metabolites at an early time point
after glucose labelling was validated as a reliable measure of PKM2
activity using an activator of PKM2 (Supplementary Figs 5 and 6 and
Supplementary Discussion).
In the cytosol, pyruvate is metabolized to lactate by lactate dehydrogenase (LDH) and the resulting lactate contains three glucose-derived
carbons. In addition, pyruvate is translocated to the mitochondria
where it is oxidized and de-carboxylated to acetyl-CoA, which enters
the tricarboxylic acid (TCA) cycle to form citrate, contributing two
carbon atoms from glucose. When cells were incubated with U-13Cglucose, both glucose-derived lactate and citrate were detected by
LC–MS (13C3-lactate and 13C2-citrate). Blocking PKM activity shifted
the metabolism of glucose away from lactate production in the cytosol
to citrate production in the mitochondria (Fig. 2 and Supplementary
Fig. 5). Heavier isotopomers of citrate (particularly 13C4-citrate) were
detected in cells incubated for a longer time period (4 h) with U-13Cglucose, an indication of further oxidation and generation of citrate in
the TCA cycle (Supplementary Fig. 7). These results are in line with the
observed increase in oxygen consumption of shPKM cells (Supplementary Fig. 3).
An increased metabolic flux into the serine and glycine biosynthetic
pathway was also observed in cells with reduced pyruvate kinase activity, as determined by the accumulation of glucose-derived 13C in
these amino acids (Fig. 2 and Supplementary Fig. 5). This is the first
1
Cancer Research UK, The Beatson Institute for Cancer Research, Switchback Road, Glasgow G61 1BD, Scotland, UK. 2Astex Pharmaceuticals, 436 Cambridge Science Park, Milton Road, Cambridge CB4
0QA, UK. 3Institute of Molecular, Cell and Systems Biology, College of Medical, Veterinary and Life Sciences, Joseph Black Building, B3.09, University of Glasgow, Glasgow G12 8QQ, Scotland, UK. 4Groningen
Bioinformatics Centre, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, Groningen 9747 AG, The Netherlands. {Present address: MRC Cancer Cell Unit, Hutchison/
MRC Research Centre, Hills Road, Cambridge, CB2 0XZ, UK.
*These authors contributed equally to this work.
0 0 M O N T H 2 0 1 2 | VO L 0 0 0 | N AT U R E | 1
©2012 Macmillan Publishers Limited. All rights reserved
RESEARCH LETTER
a
b
PKM1
Medium
Cytosol
Glucose
Glucose
PKM2
30
PKM
(mM)
(mM)
6
4
2
20
10
0
0
Actin
Serine
(mM)
sh
P
(mM)
KM
a
30,000
0.1
0
Glycine
Glycine
***
20,000
15
0.8
10
0.6
5
0
Pyruvate
4
2
3
4 5 6
Time (days)
150
7
8
9
Lactate
8
*
2
1
0
6
*
4
2
0
100
TCA
cycle
(mM)
1.5
50
15
Lactate
3
(mM)
1
(mM)
0
1.0
Citrate
(mM)
0
ATP levels
(% of shCntrlb)
0.4
0.2
0
10,000
d
0.2
0
(mM)
(mM)
HCT116 shCntrlb
HCT116 shPKMb
0.10
0.08
0.06
0.04
0.02
0
(mM)
nt
rla
b
KM
sh
P
sh
C
nt
rlb
sh
C
Fluorescence
1.0
PEP
40,000
0.3
1.5
0.5
c
Serine
0.4
2.0
10
5
**
0
shCntrlb
shPKMb
**
0.5
0.0
PK
M
b
Mitochondrion
sh
sh
C
nt
rlb
0
Figure 1 | Characterization of PKM-silenced HCT116 cells. a, PKM1 and
PKM2 protein levels were detected by western blot. Actin was used as a loading
control. b, Quantified intracellular metabolite concentrations and the uptake or
secretion of extracellular metabolites in control and shPKM HCT116 cells. For
extracellular metabolites, the dashed line indicates the initial levels in the
medium and the graph represents the levels after 24 h incubation. c, The
proliferation rate of the indicated cell lines. d, Intracellular ATP levels
normalized to protein concentration in the cell extracts. Data are mean 6 s.e.m.
of triplicates and are representative of three (two in b) independent
experiments. *P , 0.05, **P , 0.01, ***P , 0.001.
direct evidence that low pyruvate kinase activity can drive serine and
glycine biosynthesis, and demonstrates an important link between key
metabolic processes observed in cancer, namely preferential PKM2
expression, aerobic glycolysis and serine biosynthesis. The relative
contribution of glucose-derived carbons to serine and glycine is low
(Supplementary Fig. 7), which is attributable to the presence of
unlabelled serine and glycine in the growth media of the cells. Indeed,
when cells were incubated for 12 h in serine- and glycine-free media,
more than 80% of the intracellular serine and glycine was glucose
derived (Supplementary Fig. 8a). Nevertheless, de novo synthesis of
serine and glycine in cells starved of these two amino acids was insufficient to recover their steady-state levels (Fig. 3a), indicating a rapid
use of newly synthesized amino acids. Importantly, the absence of
extracellular serine and glycine had a pronounced inhibitory effect
on pyruvate kinase activity, as demonstrated by a 100% increase in
PEP and a 30% decrease in pyruvate (Fig. 3a). Akin to PKM silencing,
serine deprivation reduced cytosolic lactate production and increased
mitochondrial citrate production by ,60% (Fig. 3a). Moreover, cellular
deprivation of serine and glycine for 12 h followed by 30-min incubation with U-13C-glucose resulted in a 50% decrease in the labelled
pyruvate/PEP ratio (Fig. 3b). These data indicate that serine and glycine
deprivation decreases PKM2 activity in cells, such that more glucosederived carbon is channelled into serine and glycine biosynthesis.
PKM2 is a tightly regulated enzyme that responds not only to the
availability of PEP and ADP substrates, but also to the upstream
glycolytic metabolite fructose-1,6-bisphosphate (FBP) and to phosphorylation events2,11–14. Previous studies have proposed that pyruvate
kinase activity may be regulated by amino acids15–17. Therefore, the
ability of serine or glycine to stimulate PKM2 in cells was tested. Serine
hydroxymethyltransferase converts serine to glycine and vice versa,
and therefore cells were starved of both amino acids overnight before
30-min incubation with either serine or glycine. The cells incubated
with serine or glycine showed an increase in the intracellular levels of
the added amino acid only, and the levels of the other amino acid were
unaffected (Supplementary Fig. 8b). When serine- and glycine-starved
cells were incubated for 30 min with serine together with U-13Cglucose, intracellular pyruvate kinase activity was increased relative
to the starved cells. However, glycine did not stimulate intracellular
pyruvate kinase activity (Supplementary Fig. 8c).
In order to substantiate the results observed in cells, recombinant
human PKM2 activity was further analysed in vitro. Serine was
demonstrated to activate recombinant PKM2 with a half-maximal
activation concentration (AC50) of 1.3 mM (Fig. 4a), a level that is
within the physiological range of intracellular serine concentrations
(Fig. 1b). Isothermal titration calorimetry (ITC) was used to determine
the dissociation constant (Kd) of the PKM2–serine interaction as
0.20 mM (Fig. 4b) and the titration curve was consistent with a 1:1
PKM2–monomer:serine ratio. These results not only demonstrate
direct interactions between serine and PKM2; they also suggest that
the serine concentration required for such interactions is well within
2 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 2
©2012 Macmillan Publishers Limited. All rights reserved
LETTER RESEARCH
0
0
0
PEP
+S
G
5.0 ×
60
120
180
1.0 × 105
5.0 × 104
Glycine
8.0 × 104
P = 0.002
3 × 104
4.0 × 104
180
240
0
Pyruvate
0
120
180
1 × 104
2.0 × 106
0
1.5 × 106
240
P = 0.0053
P = 0.0011
1.0 × 106
Pyruvate
0
2.0 × 105
LDH
*
1.5 × 105
5.0 × 105
1.0 × 104
1.5 × 107
1.0 × 105
60
120
180
PDH
0
240
0
60
120
180
240
Citrate
4.0 × 105
0
5.0 × 106
Citrate
P = 0.0273
2.0 × 105
1.0 ×
***
0
3.0 × 105
TCA
cycle
Lactate
1.0 × 107
5.0 × 104
0
SG
5.0 × 105
2.5 × 105
2.0 × 104
1.0 × 106
Lactate
1.5 × 106
4.0 × 104
3.0 × 104
60
2.5 × 106
–S
G
120
Glycine
**
+S
G
60
PKM
0
4 × 104
2 × 104
0
0
PEP
5 × 104
+S
G
0
1.2 × 105
+S
G
P = 0.0009
b+
240
2.0 × 105
1.5 × 105
2.0 × 106
–S
G
4.0 × 106
2.0 × 105
104
***
0
sh
PK
M
1.0 × 105
3-PG
6.0 × 106
b+
SG
4.0 ×
**
105
sh
C
nt
rlb
6.0 × 105
P = 0.004
1.5 × 105
**
50
nt
rl
240
+S
G
180
+S
G
120
–S
G
60
Serine
–S
G
0
Growth condition
8.0 × 105
Serine
2.0 × 105
–S
G
0
1.0 × 106
(% of Cntrlb+SG)
Time (min)
5.0 × 105
100
sh
C
C-
13C-
–S
G
C-
Hexose phosphate
150
–S
G
13
12
1.0 × 106
b
12C-
Peak area
shPKMa
1.5 × 106
Glucose
Peak area
2.0 ×
a
shCntrla
106
13C -pyruvate/13C -PEP
3
3
Glucose
2.5 × 106
3 × 106
105
0
0
60
120
180
240
TCA
cycle
2 × 106
**
1 × 106
the physiological range of intracellular serine levels. In agreement with
the results observed in cells, glycine could not directly activate PKM2
in vitro (Fig. 4a). Similarly to FBP, serine lowered the Michaelis constant (Km) of PKM2 for PEP by 2.3-fold, effectively increasing
the affinity of PKM2 for this substrate (Fig. 4c). Serine was found
to be the only standard amino acid that could activate PKM2 (Supplementary Fig. 9a). A fragment-based crystallographic screen18
versus human PKM2 revealed the amino acids L-alanine, L-cysteine,
L-threonine and L-serine bound to a previously uncharacterized binding pocket on PKM2. Crystallographic soaking experiments revealed
L-serine bound to PKM2, with a single L-serine molecule bound to each
of the monomers comprising the PKM2 tetramer (Fig. 4d and Supplementary Tables 1 and 2). L-serine made multiple hydrogen-bonding
interactions with PKM2 (see Supplementary Fig. 9 and Supplementary
Discussion). Additional interactions afforded by the serine side-chain
hydroxyl group rationalize the low affinity of glycine for PKM2
and may contribute to the unique ability of serine to activate PKM2
(ref. 16). The amino acid binding pocket was vacant in crystals not
soaked with serine (Supplementary Fig. 9b).
–S
G
0
+S
G
Figure 2 | The effect of PKM silencing on glycolytic flux. Cells were
incubated with U-13C-glucose and the abundance of the main glucose-derived
metabolite isotopomers was analysed at the indicated time points. The
cumulative intensities of each labelled metabolite are presented in blue
(shCntrl) or red (shPKM). The white and black circles under each metabolite
illustrate 12C and 13C labelling, respectively. All metabolic quantifications are
presented as mean 6 s.e.m. of triplicates and are representative of three
independent experiments. 3-PG, 3-phosphoglycerate; PDH, pyruvate
dehydrogenase. P values at 4 h are shown where significant.
Figure 3 | Serine and glycine deprivation changes glucose metabolism.
a, HCT116 cells were incubated for 12 h with U-13C-glucose in the presence
(1SG) or absence (2SG) of serine and glycine. The abundance of the main
glucose-derived isotopomers (black bars) and the unlabelled fraction (white
bars) of the indicated metabolites were analysed. The circles represent the
relevant isotopomers as in Fig. 2. b, Intracellular pyruvate kinase activity is
represented as the ratio between glucose-derived (13C3-) pyruvate and PEP,
30 min after labelling with U-13C-glucose. Results were normalized to shCntrl
1 SG. All results are presented as mean 6 s.e.m. of triplicates and are
representative of two independent experiments. *P , 0.05, **P , 0.01,
***P , 0.001.
On the basis of the above observations, a recombinant PKM2(H464A)
mutant was produced and demonstrated neither direct binding to, nor
activation by, serine (Fig. 4b, e). These data confirm that the aminoacid-binding site identified in the soaking experiments is the only
amino-acid-binding site on the PKM2 protein. Despite its inability
to bind serine, the PKM2(H464A) mutant was activated by FBP in a
similar way to wild-type PKM2. Conversely, a S437Y mutant of PKM2,
which cannot bind FBP19, was activated by serine (Fig. 4e, f). These
results demonstrate that PKM2 is independently activated by either
FBP or serine, and that both molecules could contribute to PKM2 regulation in response to glucose and/or amino acid deprivation in vivo.
PKM2 regulation by serine in tumours is important as the serine concentration in the blood is ,20-fold below that of glucose, hence under
an interrupted blood supply, a considerable deficit in serine supply
0 0 M O N T H 2 0 1 2 | VO L 0 0 0 | N AT U R E | 3
©2012 Macmillan Publishers Limited. All rights reserved
20
10
1
10–10 10–8 10–6 10–4 10–2 100
–0.2
PKM2 WT
PKM2(H464A)
–0.4
–0.6
0
e
30
f
20
10
Fold-activation
(relative to no enzyme control)
Fold-activation
(relative to no enzyme control)
[Compound] (M)
H464A+FBP
H464A+Ser
1
10–10 10–8 10–6 10–4 10–2 100
[Compound] (M)
6
10
20
30
Molar ratio
5
4
3
2
1
10–10 10–8 10–6 10–4 10–2 100
[Compound] (M)
Phe 470
0.6
Arg 106
0.4
L-Ser
Arg 43
0.2
0.0
0.0
40
S437Y+FBP
S437Y+Ser
d
FBP
Ser
Ctrl
0.5
1.0
1.5
[PEP] (mM)
Asn 44
His 464
2.0
g
Fold-activation
(relative to no enzyme control)
30
V0 (RLU min–1)
40
c 0.8
b 0.0
PKM2+Gly
PKM2+FBP
PKM2+Ser
Fold-activation
(relative to no enzyme control)
a
kcal mol–1 of injectant
Fold-activation
(relative to no enzyme control)
RESEARCH LETTER
15
h
15
10
5
PKM1+FBP
PKM1+Ser
1
10–10 10–8 10–6 10–4 10–2 100
[Compound] (M)
10
PKLR+FBP
PKLR+Ser
5
1
10–10 10–8 10–6 10–4 10–2 100
[Compound] (M)
Figure 4 | Serine is an allosteric activator of PKM2. a, In vitro activity of
recombinant human PKM2 was analysed in the presence of increasing
concentrations of FBP ( ), serine (#) or glycine (m). b, Serine binding to
PKM2 wild-type (WT) or PKM2(H464A) mutant was measured by ITC. The
Kd of serine for wild-type PKM2 was measured as 200 mM. No serine binding to
PKM2(H464A) was detected. c, The initial PKM2 reaction rate (V0 (relative
luminescence units (RLU) per min)) was measured at different PEP
concentrations in the presence of 50 mM FBP ( ), 100 mM serine (#) or
vehicle (m). Km values for PEP were determined as 1.9 mM, 0.81 mM and
0.19 mM in the presence of vehicle, serine and FBP, respectively. For a and
c, results are presented as mean 6 s.e.m. of three experiments. d, 2.3 Å 2Fo – Fc
map (purple) contoured at 1s for the final, refined structure of L-serine
(orange) bound to PKM2 (green). The side chain of His 464, which was
subsequently mutated to alanine, is shown in yellow. e, f, In vitro activity of
PKM2 mutants H464A (e) and S437Y (f) was analysed in the presence of
increasing concentrations of FBP ( ) or serine (#). g, h, In vitro activity of
PKM1 (g) and PKLR (h) was analysed in the presence of increasing
concentrations of FBP ( ) or serine (#). For a and e–h, the signal was
normalized to controls containing no enzyme. The basal activity of the studied
pyruvate kinase in each panel in the presence of vehicle control is indicated by
the dotted line. For e–h, data are presented as the mean of duplicate
determinations and are representative of three independent experiments.
would occur. Serine is crucial for multiple metabolic pathways required
for cell growth and proliferation, including phospholipid, purine and
glutathione biosynthesis, as well as being a methyl source for singlecarbon metabolism. In addition, serine levels are depleted at the periphery of solid tumours, hence de novo serine biosynthesis is critical for
tumour growth20,21.
The allosteric regulation of pyruvate kinase is associated with complex structural changes11,14,22. Phosphorylation of Tyr 105, adjacent to
Arg 106 in the serine-binding pocket, has been shown to modulate
PKM2 activity and FBP binding12, and oxidation of Cys 358, which
is proximal to the serine-binding pocket, inhibits PKM2 (ref. 23)
(Supplementary Fig. 9e). This indicates that the PKM2 serine-binding
pocket, and proximal residues, constitute a key structural regulatory
node for PKM2.
The predominant isoform of pyruvate kinase in cancer cells is
PKM2; however, the serine-binding site identified here is conserved
in PKM1, PKL and PKR. Therefore, the possibility that serine may be a
universal pyruvate kinase regulator was tested in vitro. PKM1 demonstrated a high degree of basal activity in the absence of exogenous
activators, and was refractory to both FBP and serine activation
(Fig. 4g). By contrast, a recombinant protein corresponding to the
common portion of PKL and PKR (referred to as PKLR) was completely inactive in its basal state and was robustly activated by FBP.
Notably, PKLR was not activated by serine (Fig. 4h). These results
indicate that only PKM2-expressing cells can respond to changes in
serine availability and support shuttling of glucose-derived carbon into
serine biosynthesis after serine deprivation.
This work provides a new understanding of the relationship between
glucose and amino acid metabolism. Serine biosynthesis is an anabolic
pathway required for growth and proliferation24. However, it recruits
carbon away from the energy-production pathway of glucose utilization.
Here, to our knowledge, we present for the first time an elucidation of
the mechanism that tightly controls the metabolic bifurcation of
glucose-derived carbon. The control of PKM2 activity through serine
availability provides a rheostat-like mechanism. When serine is abundant, PKM2 is fully active, enabling the maximal use of glucose through
glycolysis. However, when the steady-state levels of serine drop below a
critical point, an immediate attenuation of PKM2 activity occurs. This
enables the fast shuttling of glucose-derived carbon to serine biosynthesis, compensating for the serine shortfall and enabling growth
and proliferation in the absence of these amino acids (Supplementary
Fig. 1). Finally, by activating PKM2, serine supports aerobic glycolysis
and lactate production, events that are critical for cancer cell growth and
survival.
.
.
.
.
METHODS SUMMARY
shCntrl and shPKM cells were obtained by infecting cells with control non-targeting shRNA or shPKM lentiviral particles, respectively. mRNA and protein levels of
pyruvate kinase isoforms were determined by quantitative PCR and western blot.
Cell proliferation was measured using the Alamar Blue assay. ATP levels were
measured using a bioluminescent assay kit. Oxygen-consumption rate and extracellular-acidification rate were measured using a Seahorse XF24. The quantification
of intracellular and extracellular metabolites was performed by LC–MS. Isotopomer
distribution of intracellular metabolites after labelling with U-13C-glucose
was analysed with the PeakML.Isotope.TargettedIsotopes() function of the
mzmatch.R library. In vitro PKM2 activity was measured by the quantification of
ATP production using luminescent Kinase-Glo Plus reagent. ITC experiments were
performed on a MicroCal VP-ITC isothermal titration calorimeter. His-tagged
human PKM2 was purified, concentrated, crystallized and then soaked in the
presence or absence of serine and X-ray diffraction data were collected.
Full Methods and any associated references are available in the online version of
the paper.
Received 13 November 2011; accepted 23 August 2012.
Published online 14 October 2012.
1.
Tennant, D. A., Duran, R. V. & Gottlieb, E. Targeting metabolic transformation for
cancer therapy. Nature Rev. Cancer 10, 267–277 (2010).
4 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 2
©2012 Macmillan Publishers Limited. All rights reserved
LETTER RESEARCH
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
Chaneton, B. & Gottlieb, E. Rocking cell metabolism: revised functions of the key
glycolytic regulator PKM2 in cancer. Trends Biochem. Sci. 37, 309–316 (2012).
Christofk, H. R. et al. The M2 splice isoform of pyruvate kinase is important for
cancer metabolism and tumour growth. Nature 452, 230–233 (2008).
Altenberg, B. & Greulich, K. O. Genes of glycolysis are ubiquitously overexpressed
in 24 cancer classes. Genomics 84, 1014–1020 (2004).
Mazurek, S., Boschek, C. B., Hugo, F. & Eigenbrodt, E. Pyruvate kinase type M2 and
its role in tumor growth and spreading. Semin. Cancer Biol. 15, 300–308 (2005).
Yamada, K. & Noguchi, T. Nutrient and hormonal regulation of pyruvate kinase
gene expression. Biochem. J. 337, 1–11 (1999).
Possemato, R. et al. Functional genomics reveal that the serine synthesis pathway
is essential in breast cancer. Nature 476, 346–350 (2011).
Locasale, J. W. et al. Phosphoglycerate dehydrogenase diverts glycolytic flux and
contributes to oncogenesis. Nature Genet. 43, 869–874 (2011).
Pollari, S. et al. Enhanced serine production by bone metastatic breast cancer cells
stimulates osteoclastogenesis. Breast Cancer Res. Treat. 125, 421–430 (2011).
Vander Heiden, M. G. et al. Evidence for an alternative glycolytic pathway in rapidly
proliferating cells. Science 329, 1492–1499 (2010).
Dombrauckas, J. D., Santarsiero, B. D. & Mesecar, A. D. Structural basis for tumor
pyruvate kinase M2 allosteric regulation and catalysis. Biochemistry 44,
9417–9429 (2005).
Hitosugi, T. et al. Tyrosine phosphorylation inhibits PKM2 to promote the Warburg
effect and tumor growth. Sci. Signal. 2, ra73 (2009).
Christofk, H. R., Vander Heiden, M. G., Wu, N., Asara, J. M. & Cantley, L. C. Pyruvate
kinase M2 is a phosphotyrosine-binding protein. Nature 452, 181–186 (2008).
Ashizawa, K., Willingham, M. C., Liang, C. M. & Cheng, S. Y. In vivo regulation of
monomer-tetramer conversion of pyruvate kinase subtype M2 by glucose is
mediated via fructose 1,6-bisphosphate. J. Biol. Chem. 266, 16842–16846 (1991).
Spellman, C. M. & Fottrell, P. F. Similarities between pyruvate kinase from human
placenta and tumours. FEBS Lett. 37, 281–284 (1973).
Eigenbrodt, E., Leib, S., Kramer, W., Friis, R. R. & Schoner, W. Structural and kinetic
differences between the M2 type pyruvate kinases from lung and various tumors.
Biomed. Biochim. Acta 42, S278–S282 (1983).
Ye, J. et al. Pyruvate kinase M2 promotes de novo serine synthesis to sustain
mTORC1 activity and cell proliferation. Proc. Natl Acad. Sci. USA 109, 6904–6909
(2012).
Davies, T. G. & Tickle, I. J. Fragment screening using X-ray crystallography. Top.
Curr. Chem. 317, 33–59 (2012).
Allali-Hassani, A. et al. A survey of proteins encoded by non-synonymous single
nucleotide polymorphisms reveals a significant fraction with altered stability and
activity. Biochem. J. 424, 15–26 (2009).
20. Medina, M. A., Marquez, J. & Nunez de Castro, I. Interchange of amino acids
between tumor and host. Biochem. Med. Metab. Biol. 48, 1–7 (1992).
21. Márquez, J., Sanchez-Jimenez, F., Medina, M. A., Quesada, A. R. & Nunez de Castro,
I. Nitrogen metabolism in tumor bearing mice. Arch. Biochem. Biophys. 268,
667–675 (1989).
22. Mattevi, A., Bolognesi, M. & Valentini, G. The allosteric regulation of pyruvate kinase.
FEBS Lett. 389, 15–19 (1996).
23. Anastasiou, D. et al. Inhibition of pyruvate kinase M2 by reactive oxygen species
contributes to cellular antioxidant responses. Science 334, 1278–1283 (2011).
24. de Koning, T. J. et al. L-serine in disease and development. Biochem. J. 371,
653–661 (2003).
Supplementary Information is available in the online version of the paper.
Acknowledgements The work performed at the Beatson Institute for Cancer Research
was supported by Cancer Research UK. We thank D. Sumpton for technical support
with two-dimensional gel electrophoresis and N. Thompson, N. Wallis and M. Jones for
comments provided during manuscript preparation. We would also like to thank
D. M. Sabatini for the Scramble shRNA plasmid used as a control (shCntrla) and the
Structural Genomics Consortium for providing us with the PKM2 expression plasmid
from their collection. We thank A. King for editorial work and S. Tardito for graphical
help.
Author Contributions M.O. and E.G. conceived the project and wrote the manuscript
with the help of B.C., P.H. and C.F. L.Z., B.C. and C.F. performed the LC–MS assay and
analysed the raw data. A.C. and A.J. analysed the LC–MS data and identified the
different isotopomers of each metabolite. A.C.L.M. performed the in vitro enzymatic
activity, J.E.C. performed the ITC, M.O. generated the point mutant constructs, purified
the proteins and solved the crystal structure. F.P.H. performed the LC–MS validation of
the point mutant constructs. O.D.K.M. and K.H.V. performed, analysed and discussed
the long-term serine and glycine starvation experiment. B.C. and P.H. generated and
characterized the cell lines and performed all other experiments and data analysis. All
the authors discussed the results and commented on the manuscript.
Author Information Atomic coordinates and structure factors for the PKM2 crystal
structures have been deposited in the Protein Data Bank (PDB) under accession code
4B2D. Reprints and permissions information is available at www.nature.com/reprints.
The authors declare competing financial interests: details are available in the online
version of the paper. Readers are welcome to comment on the online version of the
paper. Correspondence and requests for materials should be addressed to E.G.
([email protected]) or M.O. ([email protected]).
0 0 M O N T H 2 0 1 2 | VO L 0 0 0 | N AT U R E | 5
©2012 Macmillan Publishers Limited. All rights reserved
RESEARCH LETTER
METHODS
Cell culture. HCT116 and HT29 colon cancer cells were maintained at 37 uC and
5% CO2 in high glucose DMEM (21969-035, Invitrogen) supplemented with 10%
FBS and 2 mM L-glutamine. SW620 colon cancer cells were maintained at 37 uC
and 5% CO2 in RPMI (12633-020, Invitrogen) supplemented with 20% FBS and
2 mM L-glutamine. Stable PKM knockdown and control cell lines were cultured in
the same media containing additional 2 mg ml21 puromycin.
Stable PKM silencing. HCT116, SW620 and HT29 cells were infected with lentiviral particles containing control shRNA (shCntrlb) (sc-108080) or a pool of
three PKM shRNAs (shPKMb) (sc-62820) (59-GATCCCTGTGGCTCTAGA
CACTAATTCAAGAGATTA-GTGTCTAGAGCCACAGTTTTT-39; 59-GATC
CGTCTGGAGAAACAGCCAAATTCA-AGAGATTTGGCTGTTTCTCCAGA
CTTTTT-39; 59-GATCCGTCTGGAGAAACAGC-CAAATTCAAGAGATTTG
GCTGTTTCTCCAGACTTTTT-39) (Santa Cruz Biotechnology) according to
the manufacturer’s instructions. Infected cells were selected using 6 mg ml21 puromycin and shPKM clones were analysed for PKM1 and PKM2 expression levels
using western blot analysis and quantitative PCR (qPCR). A different set of plasmids containing PKM shRNA (shPKMa) was bought from Openbiosystems
(TRCN0000037610) (59-CCGG-GAAGGGAAAGAACATCAAGAT-CTCGAGATCTTGATGTTCTTTCCCTTC-TTTTTG-39) and (TRCN0000037611) (59CCGG-CGGGTGAACTTTGCCATGAAT-CTCGAG-ATTCATGGCAAAGTTC
ACCCG-TTTTTG-39). Scramble shRNA (shScramble) (59-CCTAAGGTTAAG
TCGCCCTCGCTCGAGCGAGGG-CGACTTAACCTTAGG-39) (Addgene plasmid no. 1864) was used as a control (shCntrla). HCT116 cells were infected with
both pLKO.1-shPKM or pLKO.1-shScramble and selected using 2 mg ml21 puromycin for 2 weeks and PKM-silenced clones were analysed for PKM1 and PKM2
expression levels using western blot analysis and qPCR.
mRNA extraction and qPCR analyses. 4 3 105 cells were plated in a 6-well plate
and were lysed after 2 days in RNeasy lysis buffer (Qiagen). Lysates were passed
through QIAshredder columns (Qiagen) and mRNA was isolated using the
RNAeasy kit following the manufacturer’s instructions. RNA was quantified
and quality-controlled using an Eppendorf biophotometer and Eppendorf single-sealed cuvettes, UVette (Eppendorf UK Limited). For PCR analyses 1 mg
mRNA was retro-transcribed into complementary DNA using High Capacity
RNA-to-cDNA (AB, Life Technologies Corporation). In brief, 0.5 mM primers,
1X Fast SYBR Green Master mix (AB, Life Technologies Corporation) and 1 ml of a
1:10 dilution of cDNA in a final volume of 20 ml were used. qPCR was performed
on the 7500 Fast Real-Time PCR System (Life Technologies Corporation) and
expression levels of the indicated genes were calculated using the DDCt method by
the appropriate function of the software using actin as calibrant. The PCR program
was: 20 s at 95 uC followed by 40 cycles of 3 s at 95 uC and 30 s at 60 uC. Finally the
melting curve was performed, which was used to confirm the presence of single
PCR products. Primers are as follows: b-actin-forward primer: 59- TCCAT
CATGAAGTGTGACGT-39; b-actin-reverse primer: 59- TACTCCTGCTT
GCTGATCCAC-39; PKM1-forward primer: 59-GAGGCAGCCATGTTCCAC39; PKM1-reverse primer: 59-TGCCAGACTCCGTCAGAACT-39; PKM2-forward primer: 59-CAGAGGCTGCCATCTACCAC-39; PKM2-reverse primer:
59-CCAGACTTGGTGAGGACGAT-39. PKL-forward primer: 59-CTGGTGATT
GTGGTGACAGG-39 PKL-reverse Primer: 59-TGGGCTGGAGAACGTAG
ACT-39 PKR-forward primer: 59-CAATTTGGCATTGAAAGTGG-39; PKRreverse primer: 59- CCTGTCACCACAATCACCAG-39.
Immunoblotting. 4 3 105 cells were plated in a 6-well plate and were lysed after
2 days in radio-immunoprecipitation assay buffer (RIPA) (150 mM sodium chloride, 1.0% NP-40, 0.5% sodium deoxycholate and 50 mM Tris, pH 8.0) supplemented with a 1:100 dilution of the protein inhibitors cocktail (Sigma). Protein
concentration was determined using the bicinchoninic acid assay (Thermoscientific)
using BSA as standard (Thermoscientific). Equal amounts of protein were loaded
into 12% SDS–PAGE gels and electrophoretically separated using Tris-glycine SDS
running buffer. After SDS–PAGE, proteins were transferred onto 0.22 mm nitrocellulose (Millipore) and probed with antibodies, all at 1:1,000 dilution in 5% nonfat milk. PKM1 antibody was custom-made by PolyPeptide Laboratories using
the following peptide sequence: CLVRASSHSTDLMEAMAMGS. The PKM2 (cat
no. 3198) and PKM (no. 3186) antibodies were purchased from Cell Signaling
Technology. The anti-actin antibody (mouse monoclonal AC-40) was purchased
from Sigma. For detection, membranes were incubated with either donkey-anti
rabbit (926 32213) or donkey-anti mouse (926 32212) secondary antibodies purchased from Licor, all at 1:1,000 dilution in TBS Tween 0.1%. The infrared scanning was performed using the Licor Odyssey scanner (channel, 800; brightness, 50;
contrast, 50; sensitivity, auto; resolution, 169.492 mm; pixel area, 0.02873; intensity,
5) and acquired using Odyssey software version 3. Images were then exported as
TIFF and cropped using Adobe Photoshop CS4.
Cell proliferation. shPKM and control HCT116, HT29 or SW620 cells were
seeded into a 96-well plate at a density of 1,000 cells per well in 200 ml of
DMEM containing 2% FBS. On days 4 to 8 after seeding, 20 ml of Alamar Blue
solution (life technologies) was added to each well measured. Cells were incubated
for 6 h and fluorescence was measured using 535 nm excitation and 590-nm
emission wavelengths.
ATP levels. 6 3 105 cells were seeded on a 6-well plate the day before the experiment. Cells were then washed twice with PBS in order to remove dead cells, and
then lysed using the ATP-release buffer (Sigma). ATP was then measured using a
luciferase-based assay according to the manufacturer’s instructions using the ATP
bioluminescent somatic cell assay kit FLASC (Sigma). Values were normalized to
the total protein content of the cell lysate as measured by BCA assay (Thermoscientific) using BSA as standard.
Measurement of oxygen-consumption rate and extracellular-acidification
rate. 3 3 104 cells were plated onto XF24 plates in DMEM (10% FBS, 2 mM
glutamine) (Seahorse Bioscience) and incubated at 37 uC, 5% CO2 overnight.
The medium was then replaced with 675 ml of unbuffered assay media (Seahorse
Bioscience) supplemented with 2 mM glutamine, 25 mM glucose and 2% FBS (pH
was adjusted to 7.4 using sodium hydroxide 0.5 mM) and cells were then placed at
37 uC in a CO2-free incubator for 30 min. Basal oxygen-consuption rate (OCR)
and extracellular-acidification rate (ECAR) were recorded using the XF24 plate
reader. At the end of the experiment 1 mM antimycin A was added in order to
measure mitochondria-independent oxygen consumption. Each measurement
cycle consisted of 3 min mixing, 3 min waiting and 4 min measuring. OCR and
ECAR were normalized to cell number. To obtain the mitochondrial-dependent
OCR, only antimycin-A-sensitive respiration was used. Homogeneous plating and
cell count were assessed by fixing the cells with 10% trichloroacetic acid for 1 h at
4 uC and then staining the fixed cells with 0.47% solution of Sulforhodamine B
(Sigma).
Quantification of intra- and extracellular metabolites by the standard addition method. 1 3 106 cells were plated onto 6-cm plates in triplicates and cultured
in standard medium (DMEM, 10% FBS, 2 mM glutamine). Two additional plates
were grown as counter plates. The medium was replaced after 24 h by 10 ml of
fresh standard medium, and cells were incubated for another 24 h before extraction (as described in the following section). Standard compounds were weighed
separately and dissolved together in water to make solution A (in which each
metabolite has a concentration between 1 mM and 10 mM). 1 ml of solution A
was added to 49 ml of dilution solvent (50:50 acetonitrile:water) to make stock
solution B (in which each metabolite had a concentration between 20 mM and
200 mM). For quantification, cells or media extracts (200 ml) were mixed with
800 ml of dilution solvent, containing 0, 4, 20, 100, 300 or 500 ml of stock solution
B. Dilutions were analysed by LC–MS. The concentration of each metabolite in the
extract was calculated according to the linear regression fit25. All dilution series
were performed in triplicates using three biological replicates.
Measurement of 13C-labelled metabolites by LC–MS. 4 3 105 cells were plated
onto 6-well plates and cultured in standard medium for 24 h. The medium was
then replaced by 2 ml of fresh medium containing 5 mM unlabelled glucose and
3 h later 5 mM of U-13C-glucose (Cambridge Isotope Laboratories) was added;
alternatively, medium was replaced by 2 ml of fresh medium with U-13C-glucose
only. Cells were incubated for the indicated time before extraction. For extraction,
cells were washed twice in PBS and metabolites were extracted on a dry ice/
methanol bath in a 50:30:20 ratio of methanol:acetonitrile:water and quickly
scraped. The insoluble material was spun down in a cooled centrifuge at
16,000g for 15 min at 0 uC and the supernatant was collected for subsequent
LC–MS analysis. The volume of extraction solution was calculated according to
cell number and, extrapolated using a ‘counter dish’ cultured under the same
experimental conditions as the sample dishes. A volume of 1 ml of extraction
solutions per 2 3 106 cells was used. Metabolites were separated using a liquid
chromatography system. A ZIC-pHILIC column (4.6 mm 3 150 mm, guard column 4.6 mm 3 10 mm; Merck) was used for liquid chromatography separation
using gradient elution with a solution of 20 mM ammonium carbonate, with 0.1%
ammonium hydroxide, and acetonitrile. Detection of metabolites was performed
using a Thermo Scientific Exactive high-resolution mass spectrometer with electrospray ionization, examining metabolites in both positive and negative ion
modes, over the mass range of 75–1,000 m/z.
2-dimensional gel electrophoresis. shPKM and control HCT116 cells were lysed
precipitated and resuspended as described in ref. 9. Isoelectric focusing (IEF) was
performed using ZOOM strips (pH 3–10, non-linear) according to the manufacturer’s instructions (Invitrogen). After IEF, strips were equilibrated in buffer containing 10 mg ml21 DTT and subsequently in buffer containing 25 mg ml21
iodoacetamide before SDS–PAGE. As negative control, shPKM HCT116 cells
were lysed and precipitated in acetone and the pH adjusted to pH 5 by the addition
of 1% acetic acid. This removed any acid-labile phosphates such as phosphohistidine. PGAM1 and its phosphorylated forms were detected by western blot
©2012 Macmillan Publishers Limited. All rights reserved
LETTER RESEARCH
using goat anti-PGAM1 (Novus) 1:1,000 and Donkey anti-Goat (Licor) 1:1,000
antibodies.
PKM2 activator. Cells were treated for 1 h with 20 mM of the commercially
available PKM2 activator 1-(3-chloro-5-trifluoromethyl-pyridin-2-yl)-1H-pyrrole2-sulfonic acid p-tolylamide26 (indicated in this letter as Cmpd1) or with vehicle as
control. Both were incubated in media containing 25 mM U-13C–glucose for the
indicated time.
In vitro measurement of pyruvate kinase activity. PKM2 was expressed and
purified as described in the ‘PKM2 X-ray crystallography’ section below. Rabbit
muscle PKM1 was obtained from Sigma; PKLR was purchased from Abcam.
Enzyme activity was measured in vitro with a coupled assay quantifying levels
of ATP using luminescent Kinase-Glo Plus reagent (Promega) as described previously in ref. 27 with some modifications. Measurements were performed in the
presence of 50 mM Tris, pH 7.5, 100 mM KCl, 10 mM MgCl2, 200 mM PEP,
200 mM ADP, 3% DMSO and either 4 nM PKM2, 4 nM PKM1 or 10 nM PKLR.
Reactions (25 ml) were incubated for 20 min on a shaker before addition of 25 ml
Kinase-Glo Plus reagent, as per the manufacturer’s instructions. Luminescence
was read with a PHERAstar (BMG LABTECH). The signal was normalized to the
no-enzyme controls. Activation curves were fitted to a four-parameter logistic
equation and Km curves were fitted to a Michaelis–Menten equation using
Prism 5 (GraphPad).
Isothermal titration calorimetry. Isothermal titration calorimetry experiments
were performed on a MicroCal VP-ITC at 25 uC in a buffer comprising 50 mM
Tris, 100 mM KCl, 10 mM MgCl2 and 1 mM Tris(carboxyethyl phosphine)
(TCEP) at pH 7.5. For titrations the L-serine concentration was 5 mM in the
injection syringe and the PKM2 concentration was 28 mM in the sample cell.
The protein concentration refers to the monomer. PKM2 was incubated for
30 min with an excess of FBP (200 mM) before the L-serine titration was performed
in the presence of FBP. The Kd value for L-serine binding was much greater than
the PKM2 concentration used, making it difficult to accurately determine the
stoichiometry value. Therefore, the stoichiometry parameter was fixed at 1 for
the purpose of data analysis using the single site-binding model in Origin 7.0.
PKM2 X-ray crystallography. A publicly available human PKM2 expression
construct was obtained from the Structural Genomics Consortium (SGC). His6PKM2 was purified using nickel-nitrilotriacetic acid affinity capture and HiLoad
Superdex 16/60 S75 size-exclusion chromatography. Human PKM2 was crystallized using hanging-drop vapour diffusion. Protein solution (10 mg ml21, 25 mM
Tris/HCl, pH 7.5, 0.1 M KCL, 5 mM MgCl2, 10% (v/v) glycerol) was mixed in a 1:1
ratio with reservoir solution containing 0.1 M KCl, 0.2 M ammonium tartrate, 24%
(w/v) PEG3350. Crystals were soaked overnight in a solution containing 30 mM
L-serine, cryo-protected and flash-frozen in liquid nitrogen. X-ray diffraction data
were collected from a single crystal at 100K at Beamline-I03 at the Diamond Light
Source. Diffraction data were processed using XDS AutoPROC from Global
Phasing and SCALA (CCP4)28. Molecular replacement was performed using
model 3H6O (SGC) in CSEARCH29 and maximum-likelihood refinement carried
out using a mixture of automated (see ref. 18 and references therein) and manual
refinement protocols employing Refmac (CCP4) and AutoBuster from Global
Phasing. Ligand fitting was performed using Autosolve29 and manual rebuilding.
Simulated annealing was not used. The four PKM2 monomers comprising the
tetramer in the asymmetric unit were refined as independent entities, but noncrystallographic symmetry (NCS) restraints were imposed in AutoBuster using the
‘ncsauto’ command. Refinement of the structure in the absence of NCS restraints
gave (Rf 5 23.7, R 5 17.7) and with ‘ncsauto’ gave (Rf 5 22.7, R 5 17.9), showing a
small, but significant, reduction in Rf using ‘ncsauto’ restraints. At the ‘effective
resolution’ of 2.36 Å there are ,86,000 unique reflections. The refinement
included ,16,600 non-hydrogen atoms. B-factors were refined isotropically giving
a total of ,66,500 parameters for all non-hydrogen atoms in the PKM2 tetramer.
The four serine molecules were refined as independent ligands. Details of ligand
occupancies, B-factors and so on are tabulated in the crystallographic data tables
(Supplementary Tables 1 and 2)30–33.
PKM2 mutagenesis. Human PKM2 point mutant constructs were generated
using a Stratagene QuikChange II site directed mutagenesis kit (no. 200524).
PCR protocols were as defined in the product manual. The following forward
DNA primers, and their reverse complemented primer counterparts, were used
for the mutagenesis reactions (sequence of mutated bases shown in uppercase
bold): H464A, 59gctcgtcaggccGCcctgtaccgtggc39; S437Y, 59accaagtctggcaggtAtgct
caccaggtgg39. Primers were purchased from Sigma. The previously described SGC
human PKM2 construct was used as the DNA template within the PCR reactions.
The presence of the point mutations was confirmed by DNA sequencing of the
DNA constructs (Beckman Coulter Genomics Inc.) and in-house LC–MS of the
purified recombinant proteins. The mutant proteins were expressed and purified
identically to the wild-type protein.
Statistical analyses. The data (mean 6 s.e.m.) are representative of three independent experiments, performed in triplicates (unless otherwise indicated). Data
were analysed and presented with GraphPad Prism 5.01 software (GraphPad
Software Inc.). P values were calculated using an unpaired two-tailed t-test.
Data processing. The data-processing workflow started by first converting the
vendor specific raw data files from the mass spectrometer into the mzXML opendata format34, using the msconvert utility from the ProteoWizard Library and
Tools collection35 (http://proteowizard.sourceforge.net/). The set of all chromatographic peaks in each of the converted raw files were then extracted using the
CentWave36 feature detection algorithm from XCMS37. The resulting data were
stored in the PeakML file format38, and the rest of the processing was handled by
the scriptable mass spectrometry data-processing tool mzmatch.R39 (http://
mzmatch.sourceforge.net/).
The next step in the workflow involved aligning and combining the chromatographic features between biological replicates of a single sample. The PeakML files
thus created were subjected to an additional filtering procedure to discard all peaks
that were not reproducibly detected in all biological replicates involved.
Chromatographic peaks of individual samples were then aligned together on the
basis of their retention time and m/z values, and combined into a single PeakML
file. Peak sets that do not include peaks from every sample were filled in by
extracting ion chromatograms within the retention time and mass window of
the corresponding peak set directly from the raw data files. From these peak sets,
only those that had more peaks than the number of replicates minus one were
selected for further analysis. Putative identification of the peak sets were made by
matching the detected masses to that of the compounds relevant to this study.
Isotope peaks were extracted by identifying the peaks that fell in the retention time
window of the identified unlabelled peak and correspond to the estimated mass
window (2 p.p.m.) of the isotope. All isotope identification and quantification of
the ratios were performed by the PeakML.Isotope.TargettedIsotopes() function of
the mzmatch.R library, and detailed documentation and tutorials for which are
available at http://mzmatch.sourceforge.net/isotopes targetted.html.
25. Luo, B., Groenke, K., Takors, R., Wandrey, C. & Oldiges, M. Simultaneous
determination of multiple intracellular metabolites in glycolysis, pentose
phosphate pathway and tricarboxylic acid cycle by liquid chromatography-mass
spectrometry. J. Chromatogr. A 1147, 153–164 (2007).
26. Salituro, F. G. & Saunders, J. O. Therapeutic compositions and related methods of
use. PTC patent application WO/2010/118063 (2010).
27. Boxer, M. B. et al. Evaluation of substituted N,N9-diarylsulfonamides as activators
of the tumor cell specific M2 isoform of pyruvate kinase. J. Med. Chem. 53,
1048–1055 (2010).
28. Collaborative Computational Project, Number 4. The CCP4 suite: programs for
protein crystallography. Acta Crystallogr. D 50, 760–763 (1994).
29. Mooij, W. T. et al. Automated protein-ligand crystallography for structure-based
drug design. Chem Med Chem 1, 827–838 (2006).
30. Brünger, A. T. Free R value: a novel statistical quantity for assessing the accuracy of
crystal structures. Nature 355, 472–475 (1992).
31. Diederichs, K. & Karplus, P. A. Improved R factors for diffraction data analysis in
macromolecular crystallography. Nature Struct. Biol. 4, 269–275 (1997).
32. Weiss, M. S. & Hilgenfeld, R. On the use of the merging R factor as a quality indicator
for X-ray data. J. Appl. Crystallogr. 30, 203–205 (1997).
33. Weiss, M. S. Global indicators of X-ray data quality. J. Appl. Crystallogr. 34, 130–135
(2001).
34. Pedrioli, P. G. et al. A common open representation of mass spectrometry data and
its application to proteomics research. Nature Biotechnol. 22, 1459–1466 (2004).
35. Kessner, D., Chambers, M., Burke, R., Agus, D. & Mallick, P. ProteoWizard: open
source software for rapid proteomics tools development. Bioinformatics 24,
2534–2536 (2008).
36. Tautenhahn, R., Bottcher, C. & Neumann, S. Highly sensitive feature detection for
high resolution LC/MS. BMC Bioinformatics 9, 504 (2008).
37. Smith, C. A., Want, E. J., O’Maille, G., Abagyan, R. & Siuzdak, G. XCMS: processing
mass spectrometry data for metabolite profiling using nonlinear peak alignment,
matching, and identification. Anal. Chem. 78, 779–787 (2006).
38. Fernandez, C. A., Des Rosiers, C., Previs, S. F., David, F. & Brunengraber, H.
Correction of 13C mass isotopomer distributions for natural stable isotope
abundance. J. Mass Spectrom. 31, 255–262 (1996).
39. Scheltema, R. A., Jankevics, A., Jansen, R. C., Swertz, M. A. & Breitling, R. PeakML/
mzMatch: a file format, Java library, R library, and tool-chain for mass
spectrometry data analysis. Anal. Chem. 83, 2786–2793 (2011).
©2012 Macmillan Publishers Limited. All rights reserved